Claim and evidence
Run fewer but genuinely different ad concepts (different message, persona, format) instead of many near-copies.
No dated example yet. Treat this tactic as a hypothesis.
Anecdotal
Practitioner experience without published numbers.
Working · 2026-02 Practitioner advice; no official Meta numbers fetched.
What the numbers show
No published number for this tactic has passed our word-for-word check yet, so there is no chart.
Why it works
Practitioners say Meta's retrieval groups similar ads, so cosmetic edits add little new information while distinct angles give the system real options.
How to run it
- Define 4-6 distinct angles by audience and motivation
- Vary the message, not only visuals
- Test in a consolidated structure
- Retire fatigued concepts and replace with new angles
- Log which angle wins by audience
Ways to run it
Design options to test, not promised results.
- Option APersona-led angles
Write one concept per buyer persona, each with its own problem and promise.
- Option BFormat mix
Carry the message across static, video, carousel and creator formats.
- Option CFatigue rotation
Retire concepts as they fatigue and replace them with new angles, not lightly edited copies.
Dated examples
No dated example yet. Treat this tactic as a hypothesis.
Does it still work?
Practitioner advice; no official Meta numbers fetched. source
Risks
- Claims about Andromeda are partly speculative
- Counts (3-5 vs 10+) are not official
Sources
- convert.com
“Under Andromeda, small tweaks don't count as new information.”
✓ quote checked word-for-word on the page - runneth.com
“Five product shots with slightly different copy variations? Meta sees that as one ad now.”
✓ quote checked word-for-word on the page - dentsu.com
“Duplicating ads with small edits is no longer enough.”
✓ quote checked word-for-word on the page
Change log
- 2026-11-20 — harvest v0.1 (acq-paid)